consumer preferences for table cassava …ageconsearch.umn.edu/bitstream/94839/2/artigo 4.pdf ·...

Carolina González & Nancy Johnson

363

ISSN 1679-1614

1 Received August 11, 2009; Accepted November 19, 2009.2 Research associate in the International Center for Tropical Agriculture (CIAT), Cali, Colombia and a PhD

candidate, Department of Agricultural Economics and Social Sciences, University of Hohenheim, Stuttgart,Germany. E-mail: [email protected] or [email protected].

3 Leader of the Livestock, Livelihoods and Poverty project in the International Livestock Research Institute(ILRI), Nairobi, Kenya. E-mail: [email protected].

CONSUMER PREFERENCES FOR TABLECASSAVA CHARACTERISTICS IN

PERNAMBUCO, BRAZIL1

Carolina González2

Nancy Johnson3

Abstract: Cassava is a major source of carbohydrate for populations in the tropics;however, there is little information about the preferences of consumers toward thequality characteristics of this crop. This paper analyzes the demand for differentcassava attributes, and applies the hedonic price method to estimate the values thatconsumers give to implicit attributes of cassava. The results show that ease of peeling,time of cooking and texture of cassava are the most important characteristics consumersconsider when purchasing and consuming cassava. Cassava varieties, root size, ease ofpeeling and location of the market are relevant attributes in price determination.

Keywords: cassava, consumer preferences, hedonic price, Northeast of Brazil

1. Introduction

Cassava (Manihot esculenta Crantz) is a root crop from tropical America.It is the fourth most important food staple produced in the tropics, with aglobal production of 228 million tons (FAOSTAT, 2008) and it is a majorsource of carbohydrate for populations in the humid tropics, around 700million people obtain more than 500 calories per day from cassavaconsumption (HARVESTPLUS, 2008). It is in Sub-Saharan Africa wherethe per capita consumption is the highest (101 kg/year) (FAOSTAT, 2008).In Latin America and the Caribbean the consumption per capita is lower

REVISTA DE ECONOMIA E AGRONEGÓCIO, VOL.7, Nº 3

364

(21kg/year) however, in Brazil cassava continues to be a principle staplefood and average per capita consumption is 41 kg/day (FAOSTAT, 2008).

Cassava is usually considered a subsistence crop, grown and eaten bythe poor. However, recent studies in Brazil suggest that because ofmigration from rural to urban zones and price seasonality among otherfactors; many people purchase their cassava, even if they also produce(SOUZA, FARIAS, MATTOS, et al., 2006). This means that attentionmust be paid to consumer and market characteristics as well as toproduction characteristics such as yield and disease resistance, whichhave previously been a major focus on breeder attention (CIAT, 2007).

There are many food products derived from cassava. The traditionalcategories according to the root type are: table or sweet cassava andindustrial or bitter cassava. In Brazil, most table cassava is distributedfor direct consumption as fresh cassava called macaxeira or aipim. Farinha(a toasted flour) and starches are the main sub products of industrialcassava. Several studies have looked at processed products of cassava,especially in farinha and starch production and commercialization(CAPRILES, SOARES and AREAS, 2007; SOUZA, FARIAS,MATTOS, et al., 2006); however, there are very few studies about freshcassava. Due to cassava’s importance in the agricultural market and dietof the poor people in the Northeast (NE) of Brazil, the objective of thispaper is to help fill the knowledge gap about consumer preferences forfresh cassava. First, using a hedonic price analysis, we attempt to measurethe consumer’s implicit price of cassava characteristics. This approachpostulates that the price of goods is a function of the quality characteristicsof the product. To complement this analysis, using logit models we lookat what consumer say about their preferences for specific cassavaattributes. These results should be useful for producers and sellers offresh cassava since they show to what extent quality differentials arereflected in price. They could also be useful for researchers in theirdecisions about characteristics to consider in crop improvement programs.The paper is organized as follows. Section 2 and 3 describe the backgroundand the theoretical model employed in the analyses. Section 4 presents


365

the data and describes the variables used in the analysis. The empiricalresults are reported and discussed in the Section 5. Finally the conclusionsare presented in Section 6.

2. Background

2.1 Crop Characteristics

Historically cassava has played a fundamental role in Brazil as source ofcarbohydrates for human consumption and as a supply of employmentand income in poorer rural areas especially in the Northeast (NE). Cassavavarieties are often categorized as either “sweet” (macaxeira or aipim) or“bitter” (mandioca4), reflecting the absence or presence, respectively, oftoxic levels of cyanogenic glucosidesis. The former can be consumeddirectly after peeling and either boiling, baking or frying, while the latterneeds additional processing such as fermentation or drying. The bittervarieties of cassava are used for industrial uses (OSPINA andCEBALLOS, 2002).

In optimal conditions5 cassava requires at least 10 months of warmweather to produce a crop. It is traditionally grown in a savanna climate,but can be grown in extremes of rainfall (O’HAIR, 1995). Certain inherentcharacteristics have made cassava an important crop in Brazil: it hasvery high productivity per unit land area; it is well adapted to adverseclimatic and soil conditions; it has no fixed planting date or time of harvest;and its rarely fails as a crop.

4 Popular Portuguese name.5 The production cycle is the same for both sweet and bitter varieties.


366

2.2 Price

Two characteristics strongly influence the price of fresh cassava—perishability and competition with other derivate cassava products. Thesefactors plus a market composed of small producers with low technologyadoption, low degree of organization and lack of access to informationlead to significant fluctuations in prices. In Brazil, studies reveled thatcassava has a demand elasticity less than 1; CARDOSO AND SOUZA(1999) showed some elasticity coefficients: -0,02 (1970), -0,02 (1975)and –0,03 (1975). In this condition incentives for more production couldbe perverse and harmful for producers, leading to reductions in pricesand, by extension, in producer incomes. For these reasons it is importantto work to add value to fresh cassava, focusing in improve the attributes,which differentiate the product in markets.

3. Theoretical Model

Much work has been done on the impact of quality characteristics onprice of agricultural products in developed countries (WAHL, SHI, andMITTELHAMMER, 1995; BOWMAN and ETHRIDG,1992). However,few empirical studies have been conducted to quantify the value of qualitycharacteristics of traditional crops in developing countries (UNNEVEHR,1986; SAMIKWA, BRORSEN and SANDERS, 1998; DALTON, 2004;EDMEADES, 2006). This method presumes that the price of a marketedgood is related to its characteristics. Therefore the observed productprice is constructed by the attributes of the product (WILLIAMS,SPYCHER and OKIKE, 2003). The marginal implicit value of outputcharacteristics can be derived from a hedonic price function that tracesthe behavior of consumers and producers of differentiated products(EDMEADES, 2006).

The buyer’s bid function is derived through utility maximization subjectto an expenditure constraint, and it can be represented by the utilityfunction.


367

),;,....,( 2,1 αµµ Yzzz m= , (1)

where, µ(.) is the buyer’s bid function for the product in the market, z isa vector of the characteristics of the good, Y is the buyer’s total expenditureand α is a vector of observed and unobserved parameters, whichcharacterize the preferences of the consumer. The fist partial derivativeof the bid function with respect to an individual characteristic depict thebuyer’s willingness to pay for an additional unit of the characteristic(CAREW, 2000).

On the supply side, the seller’s offer function can be specified as:

),;,...,( 21 γϕϕ Nzzz m= , (2)

where ϕ (.) is the seller’s offer function, N is the output quantity of goodproduced by the seller with characteristic specification z, and ã is a vectorof prices and production technologies. The offer function is defined asthe minimum price that the seller is willing to accept for supplying N unitsof good having characteristic levels z. The first partial derivative of theoffer function with respect to an individual characteristic reveals theseller’s marginal implicit valuation for providing other unit of thatcharacteristic.

In the market, the sales occur when both, buyer and seller agree on theprice of a particular product with a specific set of characteristics. Theintersection point between the buyer’s bid curve and the seller’s offercurve for the characteristics represents this situation. Simultaneously, theintersection point also represents the buyer’s and seller’s marginal implicitbid and offer, respectively (WAHL, SHI, and MITTELHAMMER, 1995).

Finally, the hedonic price function is estimated by regressing the equilibriumprice of products on the characteristics of the product. It can be expressedas:


368

),....,,()( 21 mzzzfzP = , (3)

where P (z) is the price of a good and z is a vector of quality characteristicsof the good.

4. Methodology

The NE suffers the highest levels of poverty and underdevelopment inBrazil. In this region, skewed land distribution and semi-arid climate areamong the factors that contribute to the region’s high relative levels ofinfant mortality, absolute poverty, unemployment, underemployment,illiteracy, lack of access to basic services and malnutrition (OSPINA andCEBALLOS, 2002). Pernambuco, the focus of this study, is a typicalstate in the NE Brazil. In terms of population it is second after Bahiawith an estimated population of 8.5 million in 2007 (IBGE, 2007). Over85% of the area of Pernambuco falls into the category of semi-arid (lessthan 600 mm rainfall in a year). This state is the fourth largest producerof cassava in the NE of Brazil, approximately 660 thousand ton/year.However it has the second highest per capita cassava consumption rates,125gr/day per capita after Paraiba (WORLD BANK, 1997). In semiaridPernambuco, low and variable rainfall makes cassava practically theonly staple food crop option for farmers, and cassava consequentlyconstitutes the main food source, especially for low-income people.

For analyzing consumer preferences we conducted a survey in the stateof Pernambuco during the end of 2006 and beginning of 2007. Theinterview-based survey was carried out in urban areas of four medium-sized municipalities with high production (in their rural areas) andconsumption of cassava. These municipalities represent the two majorgeographic zone of this state: a. Semiarid (Agreste and Sertao) and Coastal(Zona da Mata and part of Agreste). We took two municipalities fromsemiarid and two from coastal. They are typical cassava production zoneswith different varieties of sweet cassava in the local markets. A stratified


369

random sampling method was employed: households were selectedrandomly after stratifying each city into zones by income6. A samplecomprising 414 respondents was achieved. However 473 observationswere achieved because there are more than one purchase for somehouseholds. Personal interviews were conducted in the people’s homewith the person in charge of purchasing the household food.

4.1 Variables and empirical model

In this study we only refer to sweet fresh cassava varieties (macaxeiraor aipim), for direct consumption. We focused on this type of varietiesbecause they are very important in the diet of poor people. As mentionedearlier, very few studies have addressed looked at the market for thiscrop –a market that could potentially increase due to Brazil’s trend towardurbanization.

Based on a pilot study and on expert opinion of cassava researchers, weidentified possible quality characteristics that consumers might considerwhen buying and consuming sweet cassava. Specifically, we looked atthe following characteristics:

a. Colour: We differentiate the peel colour from the flesh colour. In thisregion consumers normally find in the market cassava with three peelcolours: white, pink and yellow. The flesh colour of pink and whitepeel varieties is white while yellow varieties have a same peel andflesh colour. In some cases, names of the varieties consumed areassociated with these colours.

b. Time of cooking: for consumers a good cassava takes around 15 – 20minutes for cooking after boiling. In the pressure cooker, it shouldonly take 5 minutes.

6 We interviewed people of medium and low income. However, it is important to say that three of the fourmunicipalities do not have people with high income. In these cases we basically included all the urban areaof the municipality.


370

c. Taste: Some cassava varieties are considered sweet, while othershave a more neutral flavor.

d. Texture: This refers to the level of hardness in chewing the cassava.The options in this study are mush or mealy.

e. Easy of peeling: It is very common when people buy cassava to takea little portion to tell if peeling is easy or not. Ease of peeling indicatescassava good quality.

f. Fiber: Cassava is considered fibrous when some strands are difficultto chew. While this is an undesirable characteristic, it is a very difficultone to detect at visually.

g. Size: This refers to the thickness (diameter) of the root. We dividedthem into fine (18 – 40 mm) medium (41- 55 mm) and thick (>55mm).

We also collected information on the price of cassava by variety, wherethe cassava was purchased, and quantity and frequency of cassavaconsumption in the household, along with demographic information (Table1).

Excluding the influence of market forces that can affect general pricelevels, an empirical model for fresh cassava can be specified as:

Price/kg= ILocationEasepeelTIME

TextureFiberTasteSizeyTypeVariet

εβββββββββ

++++++++

876

543210.

Since the model includes only dummy variables to measure qualitycharacteristics, except time of cooking, the estimated coefficientsdetermine the ranking pattern of each attribute on price.


371

5. Results

Most of the interviewees were female (93%) probably because they arethe people who did the majority of shopping in the household (Table 1).Fifty five percent only have elementary education and around a 10%were below the poverty line of US 1 a day per capita. The averagenumber of people in the household is 4.4, and 67% of households havechildren under 5 years. Approximately, 50% of respondents arehousewives and 9% have a formal employment, mostly in the publicsector. Around 80% of the respondents buy their cassava in the traditionalor local markets, a pattern that is consistent across the two regions. Inthe semiarid region, the percentage of people who buy the crop insupermarkets (10%) is higher than in the coastal region (only 1%), wheresellers who go door to door commonly sell fresh cassava.

Households consumed cassava on average 2.84 times per week, with slightlymore frequent consumption in the semiarid than the coastal zones (Table 1).This number shows the importance of cassava as a basic staple, however itis also shows that people do not consume it every day as they do with somestaples such as rice in Asia or maize in East Africa and Central America.The average quantity of cassava eaten per meal in a household is 335g.Finally, the average of amount spent on cassava per week in the household isR.$ 1.84/kg, which is approximately 2% of total food expenditure.

Regarding the preferences, respondents were asked to rank, in order ofimportance, the three main characteristics they consider when buying oreating cassava. The results show that ease of peeling (29%) is the mostimportant characteristic for consumers. One possible explanation is thatthis characteristic is easy to test, and people consider it an indicator notonly of amount of work involved in peeling but also of other qualitycharacteristics. After ease of peeling, time of cooking (28%) is anotherimportant characteristics for consumers, followed distantly by texture(16%) and then colour (11%). Price has the lowest place in the consumerranking, which is consistent with the price inelasticity of cassava.


372

We have a special interest in colour preferences because there are cassavavarieties with other colours different from the commonly white that couldbe introduced to markets as a high value product (HARVESTPLUS, 2008).The consumers of the areas that we study distinguish between two colours:white and yellow. The most popular varieties have the former colour. Thelatter colour is better known in the interior (semiarid region) of the country,where people called these types of varieties manteguinha, which meansbutter in Portuguese. In the semiarid region people consume more yellowcassava than coastal, 50% versus 17%. We asked about reasons whyconsumers do not purchase or consume yellow varieties. In the semiarid,they mentioned that manteguinha does not cook very well or takes moretime than white cassava (12%) for cooking7. On the coast, the main reasonsare that they have never tasted, eaten or seen it (75%).

5.1 Hedonic price

In the literature, there is some debate regarding the most appropriatefunctional form to use to estimate the hedonic function. In general, thetheory underlying the approach does not provide much guidance aboutwhich of these functional forms is most appropriate. ROSEN’s (1974)work suggests that hedonic function not be linear (CROPPER, DECKand MCCONNELL, 1988). In this study we used the Akaike InformationCriterion (AIC)8 to select the functional form of hedonic price model;we tested linear, semilog, double-log, quadratic, and a Box-Coxtransformation technique. According to the AIC test, linear and semilogwere the best functional forms. However, because semilog form hasadditional properties, it was selected as useful choice for hedonic pricemodel9. Price flexibilities—defined as the percentage of change in the

7 The results suggest that it is a wrong consumer perception; we did not find a significant difference in timecooking between white or yellow cassava (12,5 minutes).

8 This criterion minimized over choices of the number of parameters (x) in the model to form a tradeoff betweenthe fit of the model and the model’s complexity. Given a data set, competing models may be ranked accordingto their AIC, with the one having the lowest AIC being the best (EDMEADES, 2006).

9 First, the implicit value of crop characteristics may be a function not only the level of the characteristic itself,but also a function of the levels of other characteristics embodied in the crop. Semilog hedonic model areconsistent with this observation (WAHL, SHI, and MITTELHAMMER, 1995). Second, it is more useful tocalculate results expressed in price flexibilities.


373

price with respect to a 1% increase in the characteristic—were estimatedto measure sensitivities. For discrete characteristics, the price flexibilityis defined as the percentage change in the price due to the presence ofthe characteristic relative to its absence. Given the semilog specificationof the hedonic price model, marginal value has to be estimated10; it can

be expressed as 1)( −= iep β , where p is calculated at mean of

continuous variables and at zero for discrete characteristics.

Parameters obtained via estimation of semilog model, marginal valueand price flexibilities are reported in Table 2. In general, estimatedparameters were consistent with hypothesized signs, and the F test isstatistically significant. The results of the model indicate that varietieswith yellow peel colour have a higher value to consumers than pink, witha price premium of R.0.09/kg. The price flexibility of yellow varietiesshows that a presence of this characteristic, holding all else constant,increases by 11% the cassava price. Unsurprisingly, people pay for biggersizes, if cassava size decrease from thick size to medium or fine,respectively, the cassava price would reduce by 7.4 and 13% respectively.If marginal cost of changing from pink to yellow peel varieties, or ofproducing bigger cassava roots were less than R.0.08/kg and R. 0.09/kgrespectively, these results suggest that it would be beneficial for theproducer to do so. Additionally, cost of production does not depend onthe type of cassava variety; it is more related with the production system.Therefore producing cassavas with characteristics more attractive tothe market should not imply an increase in cost. The great marginal gainfor producer, however, may be associated with the location. In coastalareas cassava price is lower than in the semiarid regions. The pricedifference it is around R. 0,39/kg. Regarding to price flexibilities of dummylocation variables (semiarid), the coefficient reported is positive; holdingall else constant, cassava price in the semiarid would increase by 50%.Nevertheless, it could be not profitable for fresh cassava producers in

10 Marginal value is defined as the change in the price with respect to one unit increase in the characteristics fromits mean value. In the case of discrete variables, it depends on the presence or absence of the characteristics.(WAHL, SHI and MITTELHAMMER, 1995).


374

coastal region to distribute their production in the semiarid due to theperish ability and the high costs of transportation and refrigeration.

Unexpectedly ease of peeling has a negative coefficient, which could beattributed to the low percentage (3.8%) of consumers of hard-to-peelcassava, who pay a significantly higher price, as compared with consumersof easy-to-peel cassava. When the characteristics of consumers of hard-to-peel cassava were analyzed, results indicated that many of themproduced their own cassava or purchased cassava in supermarkets or attheir door, which could mean that these consumers pay a higher price forthis cassava because they assume they are purchasing a quality product.However, the quality of the cassava is not always as expected. Otherattributes such as texture, taste, quantity of fibers and time of cookingare statistically unimportant in terms of their influence on price. Thismay be due to the fact that these characteristics can only be known aftercassava has been boiled and consumed. This suggests thatcomplementary research needs to be done using other methodologiessuch as sensorial techniques, for example, in order to know the realeconomic importance of these characteristics.11

5.2 Logit model: Consumer preferences

We estimated a logit model for each characteristic to assess the influenceof socioeconomic and demographics characteristics of consumers oncassava preferences. Specific factors considered included sex, age,education, monthly household income, region (semiarid and coastal), adummy variable for purchase and for own production12 and kilograms offresh cassava consumed in the household per meal. For each attribute,the dependent variable was one for households that ranked that attributemost important. The estimated results show that not all the characteristicshad statistically significant models. Price, colour of the cassava, amount

11 There are some studies to attempt relate the sensory qualities of cassava roots to their physicochemical properties(PADONOU, MESTRES, and MATHURIN, 2005; BELÉIA, PRUDENCIO-FERREIRA, YAMASHITA et al.,2004), however they did not develop a economic valuation.

12 In some cases a person could be producer and buyer of cassava at the same time.


375

of fibers, and other characteristics had less than 10 % of people rankingthem as most important. However, we obtained interesting results withtime of cooking, texture, taste and ease of peeling (Table 3).

Ease of peeling is more important for women than for men, which makesense because in this zone women not only purchase but also preparethe cassava. Men, on the other hand, place more importance on taste, asignificant percentage (70%) prefer sweet cassava toward 13% ofwomen, who prefer a neutral taste.

Some studies show that there are significant differences betweenconsumer preferences among regions (SOUZA, FARIAS, MATTOS, etal., 2006), and the results of the study confirm this. While taste is veryimportant in the semiarid region, texture is determinant in the coastalregion; for example, consumers in the former region would prefer thetaste of sweet cassava while those in the latter region would prefer cassavawith a mush texture. For producers and researchers these findings arevery important to know which varieties should be produced and orientedto which regional market.

In the literature, cassava is considered an inferior good, meaning that atlower levels of income more quantity of cassava consumed in households.The results of this study support that contention; households in the studywith less income consume more grams of cassava than household withmore incomes (362 gr./meal vs. 249gr/meal). The propensity to prefertime of cooking is also negatively influenced by income. Household withhigher levels of income in the sample have lower probability of selecttime of cooking as the most important characteristics for buying andconsuming cassava, which might reflect that they are less concernedwith the fuel-related costs associated with longer cooking.13. Levels ofeducation have significant effects on which characteristics are moreimportant. Time of cooking is more important for consumers with auniversity education as compared with consumers with only elementary

13 The average per capita income is 166 reais per month (US $78). Ten percent of the households are extremelypoor; many more can be classified as moderately poor.


376

schooling, who in turn prefer ease of peeling. The preference of theformer could be attributed to the less time they have to dedicate tohousehold chores. Households with higher consumption of cassava permeal likely take more into account texture and taste of this crop atpurchase or consumption moment. Time of cooking it is less importantfor them. Finally, ease of peeling is an important characteristic for bothconsumers who have their own production and also for those who purchasecassava. These results are consistent with the raking of most importantcharacteristics that consumer consider when buying cassava. In sum,the empirical applications of the qualitative model offer valuable insightsinto the factors that influence decisions regarding the desirablecharacteristics for consumers.

6. Conclusions

This paper evaluates the consumer preferences for cassava inPernambuco, a state in NE Brazil. The aim of the paper was to fill aninformation gap about consumer preferences for quality characteristicsin order to help producers and researchers to develop varieties moreattractive for the markets. Knowledge about implicit values of qualitycharacteristics indicates which attributes should be focused on and whichcharacteristics could be allowed to vary. The empirical results presentedabove indicate that some attributes are very important when people buycassava such as ease of peeling, or time of cooking and texture for cassavaconsumption. The estimated results in the hedonic model in terms ofprices show a big difference between semiarid and coastal region, alsoamong yellow and pink varieties. The price of yellow cassava is higherthan other varieties; but its market is smaller because it is only knownand preferred in the semiarid region. Fresh cassava with larger size hasa premium. According to researchers the size of cassava depends onproduction system and environmental characteristics. Thereforeproducers have to take in account those variables in order to obtain adesirable size cassava root.


377

Consumer preferences toward characteristics such as texture and tasteare also highly influenced by region. This result suggests that producersin semiarid should grow a meal fresh cassava, with a sweet taste; whileproducers in coastal could have good market opportunities with a mushcassava, with neutral taste. Although price is relatively unimportant inthe consumers’ ranking of attributes, it is truly relevant for producers.Because of inelasticity of cassava demand, it is very important to addvalue to this crop, to avoid driving down incomes in the long term.

Finally, complementary studies should be carried out including sensorialtechniques of cassava characteristics related with an economic valuation.It is important to deepen the market study of the basic staple crops. Thistype of research could be conducted for other crops as potatoes or beansto guide producers and researchers to varieties which are most valuedby consumer.

References

BELÉIA, A., PRUDENCIO-FERREIRA, S.H., YAMASHITA, F.,SAKAMOTO, T.M.; ITO, L. Sensory and instrumental texture analysisof cassava (Manihot Esculenta, Crantz) roots. Journal of TextureStudies, v. 35, n. 5, p. 542–553, 2004.

BOWMAN, K.; ETHRIDGE, D. Characteristic Supplies and Demandsin a Hedonic Framework: U.S. Market for Cotton Fiber Attributes.American Journal of Agricultural Economics, v. 74, n. 4, p. 991-1002, 1992.

CAPRILES, V., SOARES, R.; AREAS, J. Development and assessmentof acceptability and nutritional properties of a light snack. CiênciaTecnologia de Alimentos, v. 27, n. 3, p. 562-566, 2007.

CARDOSO, C.E.; SOUZA, J. Aspectos agro-econômicos da culturada mandioca: potencialidades e limitações. Cruz das Almas: Embrapa-CNPMF, Documento 86, 1999.


378

CAREW, R. Hedonic analysis of apple prices and product qualitycharacteristics in British Columbia. Canadian Journal of AgriculturalEconomics, v. 48, n. 3, p. 241–257, 2000.

CIAT-CENTRO INTERNACIONAL DE AGRICULTURATROPICAL. Annual report 2007: Improved Cassava for DevelopingWord. Cali, Colombia, 2008. Available online at: http://www.ciat.cgiar.org/yuca/inicio.htm [Accessed 30 Sept. 2009].

CROPPER, M., DECK, L.; MCCONNELL, K. On the choice offunctional form for hedonic price functions. The Review of Economicsand Statistics, v. 70, n. 4, p. 668-675, 1988.

DALTON, T. A Household hedonic model of rice traits: Economic valuesfrom farmers in West Africa. Agricultural Economics, v. 31, n. 2-3, p.149–159, 2004.

EDMEADES, S. A Hedonic Approach to Estimating the Supply of VarietyAttributes of a Subsistence Crop. International Food Policy ResearchInstitute – IFPRI. EPT. Washington, USA: Discussion Paper 148, 2006.Available online at: http://www.ifpri.org/sites/default/files/publications/eptdp148.pdf [Accessed 30 Sept. 2009].

FAOSTAT. Database. 2008. Available online at: http://faostat.fao.org[Accessed 30 Sept. 2009].

HARVESTPLUS. Target crops: CASSAVA. 2008. Available online at:http://www.harvestplus.org [Accessed 30 Sept. 2009].

IBGE - INSTITUTO BRASILEIRO DE GEOGRAFIA EESTADISTICA. 2007. Available online at http://www.ibge.gov.br[Accessed 30 Sept. 2009].

O’HAIR, S. Cassava. Tropical Research and Education Center.Homestead: University of Florida, 1995. Available online at: http://w w w . h o r t . p u r d u e . e d u / n e w c r o p / C r o p F a c t S h e e t s /


379

cassava.html#Commercial%20Seed%20Sources [Accessed 30 June2008].

OSPINA, B.; CEBALLOS, H. La Yuca en el Tercer Milenio: SistemasModernos de Producción Procesamiento,Utilización y Comercialización.Cali, Colombia: CIAT Publications, 2002.

PADONOU, W., MESTRES, C.; NAGO, M.C. The quality of boiledcassava roots: instrumental characterization and relationship withphysicochemical properties and sensorial properties. Food Chemistry,v. 89, n.2, p. 261–270, 2005.

ROSEN, S. Hedonic Prices and Implicit Markets: Product Differentiationin Pure Competition. Journal of Political Economy, v. 82, n. 1, p. 34-55, 1974.

SAMIKWA D. BRORSEN B.W.; SANDERS L.D. Hedonic prices ofMalawi burley tobacco. International Food and AgribusinessManagement Review, v. 1, n. 1, p. 107-117, 1998.

SOUZA, L.S., FARIAS, A.R., MATTOS, P.L.; GONCALVES, W.M.(Org.). Aspectos Socioeconômicos e Agronômicos da Mandioca.1 ed. Cruz das Almas: Embrapa Mandioca e Fruticultura Tropical, 2006.

UNNEVEHR, L.J. Consumer Demand for Rice Grain Quality andReturns to Research for Quality Improvement in Southeast Asia.American Journal of Agricultural Economics, v. 68, n. 3, p. 634-641,1986.

WAHL, T.I., SHI, H.; MITTELHAMMER, R.C. A Hedonic priceanalysis of quality characteristics of Japanese wagyu beef. Agribusiness,v. 11, n. 1, p. 35-44, 1995.

WILLIAMS, T., SPYCHER, B.; OKIKE, I. Economic, institutional andpolicy constraints to livestock marketing and trade in Wets Africa.International Livestock Research Institute – ILRI, 2003. Available online


380

at: http://www.ilri.org/ilripubaware/uploaded%20files/TS_ 51011_001_West%20Africa_Marketing%20Study%20Report.pdf [Accessed 30Sept. 2009].

WORLD BANK. Living Standards Management Study (LSMS) for Brazil1996-1997. The World Bank, Washington, D.C., 1997. Available onlineat: http://www.worldbank.org/LSMS/index.htm [Accessed 30 Sept.2009].

Resumo: A mandioca-de mesa (macaxeira ou aipim) é uma das principais fontes decarboidrato das populações nos trópicos. No entanto, há pouca informação sobre aspreferências dos consumidores em relação às características de qualidade deste cultivo.Este artigo analisa a demanda de diferentes atributos da mandioca de mesa e aplica ométodo dos preços hedônicos para estimar o valor que os consumidores dão aos atributosimplícitos deste tubérculo. Os resultados mostram que a facilidade de descascamento, otempo de cozimento e a textura da mandioca de mesa são as características maisimportantes para os consumidores quando a compram e consomem. O tipo de variedades,o tamanho da raiz, a facilidade de descascamento e a localização do mercado são atributosrelevantes na determinação dos preços.

Palavras - chave: mandioca mansa, preferências do consumidor, preços hedônicos,Nordeste do Brasil


381

Table 1 - Descriptive statistics

N:414; *,**,*** The difference between semiarid and coastal is statisticallysignificant at the 0.10, 0.05 and 0.01 level, respectively

Variables Semiarid Coastal Total

Female respondent, dummy (%) 90.48 96.17 93.00**

Age of respondent (years) 39.43

(14.14) 41.31

(13.63) 40.48

(13.93)

Size of household (people) 4.45

(1.99) 4.33

(1.69) 4.40

(1.87)

No formal education 16.45 15.30 15.94

Elementary school (1 –6 years)

53.55 57.14 55.56

High school (7 – 11 years) 22.51 26.78 24.40

Edu

cati

on o

f re

spon

dent

s (%

)

University 3.90 4.37 4.11

Monthly household income (Reais) 429.45 (207.3)

395.76 (192.4)

414.55 *

(201.3)

Price (Reais/kg) .91

(.29) .59

(.18) .77 ***

(.29)

Cassava consumption (kg/meal)

.391 (0.05)

.290 (0.06)

.335 *** (.05)

# times eat cassava/ week 3.00 (1.35)

2.63 (1.32)

2.84 ***

(1.35) Amount spend/week 2.09

(1.36) 1.52 (.77)

1.84 ***

(1.17)

Con

sum

ptio

n of

cas

sava

in t

he

hous

ehol

d

Cassava producer, dummy (%)

16.45 15.85 16.18


382

Table 2 - Hedonic price of cassava

N:473; *,**,*** Statistically significant at the 0.10, 0.05 and 0.01 level,respectively

Variables Coef. Price flexibilities

Marginal Value

white -.018 (.029)

-0.018 -0.014 Variety (ref. pink)

yellow .107 ** (.047)

0.113 0.088

fine -.140 *** (.050)

-0.131 -0.102 Size (ref. thick)

medium -.077 ** (.035)

-0.074 -0.058

Taste ( ref. neutral) sweet .041 (.032)

0.042 0.033

Fibers (ref. low) much .029 (.064)

0.029 0.023

Texture (ref. mealy) mush -.021 (.031)

-0.021 -0.016

Ease of peeling (ref. Not important) -.136 ** (0.67) -0.127 -0.099

Time of cooking (min) -.003 (0.003)

-0.038 -0.002

Location (semiarid) .408 *** (.030)

0.504 0.392

Intercept -.336*** (.090)

F( 10, 462) 30.20*** R-squared 40%


383

Table 3 - Socioeconomic factors explaining cassava preferences (logitmodels)

N:414; *,**,*** Statistically significant at the 0.10, 0.05 and 0.01 level,respectively

a. Variable dropped because there are not people that simultaneously are producerand buyer.

b. Variable education-university = 0 predicts failure perfectly; it was droppedand 17 observations not used.

Time of cooking Texture Taste Ease peeling Variables

Coef. SE Coef. SE Coef. SE Coef. SE.

Female respondent .03 .44 .63 1.06 -1.66*** .62 .95** .47

Age (years) .01 .01 -.01 .02 .01 .021 .01 01

No education -.31 .33 .071 .59 .34 .78 .08 .31 High school -.11 .28 .66 .41 .94 .62 -.10 .27

Edu

cati

on

Ele

men

tary

sc

hool

University 1.92*** .61 -.34 1.09 b. -1.47** .70

Household income (R.) -.00*** .00 .00 .00 .00 .00 .00** .00

Kg/meal/per capita -8.92*** 2.45 5.54* 2.94 6.99* 1.08 2.20 1.97

Own production, dummy -.68 .50 a. a. 1.10*** .43 Cassava buyers, dummy -.29 .61 -.54 .54 .23 1.08 1.29** .54

Semiarid, dummy -.28 .23 -1.36*** .41 1.01* .61 .28 .22

Intercept 1.15 .89 -2.41* 1.37 -4.36*** 1.65 -3.81*** .87

Log likelihood -247.90 -114.27 -61.93 -264.31 Chi-squared 37.72*** 25.37*** 16.53** 24.69***


384

consumer preferences for table cassava …ageconsearch.umn.edu/bitstream/94839/2/artigo 4.pdf ·...

Documents